The discipline of AI-assisted engineering

Durable Code

Written by Pavan Kanaparthy, a Claude Certified Architect with decades of software engineering experience and years building production AI-enabled applications.

Part 1

The Premise

  1. 01 Can AI Generate Durable Code? Yes, with patience and discipline. Durable code comes down to four qualities, and to what the engineer brings: depth, breadth, and scale.

Part 2

The Discipline

  1. 02 The Three Shifts That AI Created AI changed three things about building software: who does the work, how fast decisions pile up, and how much a small team can take on.
  2. 03 The Four Qualities of Durable Code The four qualities of durable code: useful, maintainable, debuggable, and evolvable. None is a box you tick; each is a direction you keep working in.
  3. 04 The Five Failure Modes The aim here is diagnosis: five patterns seen across AI-assisted teams, named clearly enough that you can spot them in your own work.
  4. 05 The Six Practices The discipline, distilled into six practices: Specs, Manage Agents, Automated Tests, Automated Deployments, Agent Context, and Probabilistic to Deterministic.

Part 3

Claude Code

  1. 06 Claude Code: A Crash Course Treat Claude Code as a black box and the practices become superstition. A short tour of what the tool is actually doing underneath: context, memory, hooks, subagents.
  2. 07 Claude-Specific Practices The tool-specific, current-as-of-writing layer: the exact file paths, settings keys, frontmatter, and the feature catalog for Claude Code as it stands today.

Part 4

In Practice

In progress. A full repository walked through one practice at a time.